As the college football season gets closer we have decided to stick to Notre Dame’s most important matchups for this series so we can go full steam ahead on preparing content more relevant to the 2020 season. As such, we have decided to cut Bowling Green, Duke, Navy, and Boston College recaps from this series in order to focus on the closer games and the Tommy Rees coached bowl game. Please let us know in the comments if you were really looking forward to any of these recaps so adjustments can be made, and as always please follow this link to the Louisville recap for a primer on the statistics being used in this analysis.
After dismantling Bowling Green the Irish played host to the USC Trojans for a rivalry matchup in South Bend. The Irish were a 10.5 point favorite, but USC refused to go down easy. The Trojans took advantage of having speed all over the field on offense to create problems for the Notre Dame Fighting Irish, but Notre Dame was able to pull out a 30-27 win to keep the stakes high for a Michigan game that definitely was not going to be a soul crushing letdown defeat that I still think about almost every week. Anyways, let’s begin with the win probability chart.
This game featured some huge plays that really turned the tide of the game, although the model has some bugs that are creating weird spikes and valleys that should not be there. After some touchdowns you can see that there is a spike and immediate crash in win probability. It appears the model is sometimes assuming the team that scores still has possession of the ball, so as soon as the kickoff occurs their odds of winning come back down to a reasonable level. For example, the Slovis touchdown on a first and goal should not have added thirty-six percentage points to USC’s chances, because they would have had three more chances to score from close range if it had not been successful. After accounting for the valley it’s really about a fifteen percentage point increase in USC win probability, which is far more logical. We’re extremely grateful to cfbscrapr for building this model out, but not all models are perfect or above criticism.
Ignoring these weird peaks, Notre Dame did an excellent job responding to a lackluster second quarter with a second quarter that put them in control of the game. Braden Lenzy’s blazing speed added nineteen percentage points of win probability to Notre Dame’s cause, a level of high win probability they would maintain through most of the third quarter. The Trojans mounted a valiant comeback effort, but the hole they dug themselves was too deep facing an Irish team that suddenly featured a lethal ground game to run the clock out late.
After weeks of dragging on the Irish run game it is time to give those players some credit. For the first time all season running was a more efficient option than passing by EPA/play, also having a higher success rate and generating more first downs. Tony Jones led the way, getting twenty five carries and averaging an excellent 7.0 Yards and 0.40 EPA per play in one of his best performances of the year. Outside of one costly carry by Jafar Armstrong, Jahmir Smith was the only other running back to touch the ball and he also did well in limited action.
While Ian Book struggled on designed runs, he was excellent scrambling after weak performances against Georgia and Virginia. Through the air, the Irish were once again abysmal throwing behind the line of scrimmage. Each of the team’s first four games against power five conference competition they lost EPA throwing behind the line, likely because of a lack of explosive players to get the ball to on screen plays. While Chris Finke and Lawrence Keys are solid short route runners and possession receivers, they lacked the ability in open space to turn these screens and swing passes into expected points. This could be an opportunity for Braden Lenzy to increase his workload in the 2020 campaign if he’s able to stay healthy for the duration of the season.
Cole Kmet was once again spectacularly efficient and led the team in receptions. Kmet proved over and over again why he was one of the best receiving tight ends in the country, which is especially interesting knowing that I was about as involved in the Irish passing game from the press box as Kmet was the next week against Michigan. Aside from Michael Young, in what would be his final game in an Irish uniform, the receiving corps had an excellent game with everyone turning in a highly efficient performance.
Amon-Ra St. Brown was just ridiculous in this game. No one on either team had as many catches or generated as much EPA as Brown did. Running back Markese Stepp did not generate as many points as Jones but was more efficient on fifteen fewer touches. I could go on and on here because every USC player that touched the ball generated positive EPA, an incredibly remarkable feat for a team that only came away with twenty-seven points.
USC being so efficient offensively and only coming away with twenty-seven points is a testament to how important field position and taking care of the football are to winning. The Irish had no turnovers and the Trojans started no drives in Irish territory, meaning more EPA could be gained without actually putting points on the scoreboard, e.g. the bend don’t break defense. The Irish ran the ball extremely effectively and limited costly mistakes, which created a situation where USC’s skill players could all have excellent days but the team did not have a massive scoring day to show for it.
Another interesting development from this game was a two tight end set generating positive EPA! The 12 formation finally started to work a bit, with Tommy Tremble and Kmet combining to create headaches for the USC defense. Despite the success, Cole Kmet out wide in the 10 set remained the most effective way to use him. The 11 set with him playing more traditional tight end was also useful, but putting him in the backfield to create the 20 look did not work nearly as well. Kmet is a solid blocker, but he is better out wide. He was too physical for college defensive backs and too fast for college linebackers to cover, with a great ability to find creases in zone coverage. The Irish coaches did split Kmet wide a fair amount, but if I had been running the team Kmet would have lined up in the slot even more often and both of us would be wearing matching CFP championship rings as a result.
Play action consistently did not work for Notre Dame early last year. The Irish were more effective on non play action plays than play action plays in each of their four power five matchups to this point in the season, This is an interesting development because, at least at the NFL level, play action passing is much more efficient than non-play action plays. Part of the explanation could be due to the added time in the backfield creates pressured dropbacks, and Chip Long was notorious for his use of slow developing run plays and RPOs. The Irish will need to find a way to make their play action game more effective under Tommy Rees, although that likely starts with creating a more effective run game that defenses feel they have to commit to stopping.